This is a story of grief and struggle. Last Monday, I decided to power down the big editing workstation at the end of the day, rather than leave it running overnight, as I usually did for the past 12 months since I built it. It's really amazing, but leaving a few workstations running 24/7 results in a $500+ monthly tab for electricity!
So the next morning, I power up the workstation and as Windows gets to the desktop, the machine blue screens. My first reaction is "hmm.. that's odd.." and I note the error and press reset. When the machine does this a second time, I become very concerned. So now I'm watching the POST screens like a hawk. When it gets to the RAID BIOS and device detection, it's apparent that the BIOS is not loading and the first outward thing I noticed was that my LG GGW H20L Blu-ray burner was flashing a sequence of five blinks, repeating indefinately.
My next step is to troubleshoot the CMOS. I try resetting it, and I find that the machine will boot properly, but it's only running at 2.4GHz, not the usual 3.5GHz that it had been set at all year long. So I go turn on CPU Host Clock Control once more and save and reboot. At POST, the RAID BIOS fails to load again, which has me scratching my head and wondering just what is going on. Flakey motherboard?
I'll spare you the sordid details, but suffice to say that I must have cleared CMOS 20 times over the next two days and rebooted the machine hundreds of times. I also flashed the BIOS four times with different revision levels of BIOS, hoping that maybe doing so would correct a possibly-corrupt BIOS. No joy there.
The machine was acting very erraticly. When I'd save and exit the CMOS, the machne would power down. Then it finally got to the point where it would cycle the power off/on/off/on/off and stay off. I could not turn the machine back on for more than 1/2 second before it would shut itself off.
So, after removing all peripherals, testing and resetting, I concluded it was not the fault of an external peripheral. Off to the shop workbench it went. There, I could hook up another Seasonic power supply that I'd yanked from an older Win 2000 machine. Although the GeForce 8800 GTS emitted a loud sonalert tone because no ATX power connection was available from the older power supply, I could hear the beep of the motherboard booting through POST and the system remained on.
Back to the original PSU, and out comes the digital VOM for a check of voltage rails. +12... normal. +3.3... normal. +5 --only 4.2V! Ahah! Maybe this is the culprit! So I make a judgement call on what's needed in order to make the older Seasonic S12 series supply work with the new system, at least temporarily while the original unit goes for RMA repair.
The obstacle was lack of a compatible connector for the 8800GTS card. It needed a 6-pin connector, carrying 12 volts. The replacement supply loaned from the Win 2K machine has a 4-pin cable. And the keys are different, so it would not plug into the 8800. That's when I got creative with a razor blade and a needle tipped soldering iron, which I used to shave the corners of the plug to match the socket on the 8800. Lo and behold, after 10 minutes of trial and error, I was able to get the connector far enough into the socket that it worked.
The next obstacle was the supplemental 12V power jack on the motherboard, an 8-pin connector. This older PSU has shorter cables and the 12V cable was 7 inches short of the target. So I set to work splicing in 7" lengths of black and yellow #18 wires, and shrink tubing my special butt-splices which don't make a bump in the wires. That took part of an hour, but I finally got the supply hooked into the system. I was fortunate to have purchased a standard power adapter cable that takes one AT power connector and splits it to four SATA power connectors for the hard drives. I could hook everything up at last.
Back to the installation site, I plugged everything back in and booted up. So far, so good. Loaded the high performance settings and rebooted. No good. So I have a power supply problem AND a motherboard problem. Ugh! And only a year old, to boot.
The search goes on for a replacement GA-P35-DQ6 mainboard. It seems most of the online retailers no longer have it. So I hit eBay, and there is but one board for sale. But it looks like it was destined for me. The seller seems knowledgeable, has a good reason for selling it and confirmed that it ran fine at the kind of clock speeds I was used to. So I bought it and patiently await it's arrival, though that won't be til late next week, with the 3-day weekend ahead.
Meanwhile, I have to choose between limping along at a paltry 2.4GHz and lose the ability to edit multicamera timelines, or lose access to the RAID arrays and Blu-ray burner. And I have no certainty that the motherboard isn't going to go full failure in a few hours, days, or weeks.
I wonder if this board was a casualty of the new RohS compliance, which in the EDN articles I read last year, is having some serious issues with solder joints bridging over after a few months of operation. It seems that the lead-free solder is not reliable and without lead to prevent dendrite growth, fine pitch surface mount chips experience shorting after just weeks of operation in some cases. I sometimes wonder if that may play a role here, as my older Gigabyte motherboards have been great for 7 years, with the only problem being some failed electrolytic capacitors, which I was able to replace in my shop, thus saving a GA-7DXR motherboard from the scrap heap.
RohS is a real can o' worms and a disaster waiting to happen, according to the journals on electronics that I read from time to time. I looked at a number of electron micrographs of "dendrites" formed across close-pitch VLSI chip pins and remembered feeling concerned about future electronics purchases and the potential for new electronics to have a greatly-reduced operating lifespan. Whatever killed my workstation's motherboard, I just hope that it doesn't happen with the replacement that I have coming next week.
So there you have it: two whole days spent chasing down very ellusive and erratic behaviors of the workstation, 18-20-hour days spent rebooting ad infinatum, with only the worst news as a result of confirming the source of trouble. And it's not over yet. There's no certainty that the replacement "used" motherboard will work properly (although the seller assures me it ran perfectly for him at all times).
I hope it sells, so I can finance the replacement parts.
